content-add-internal-links-to-article
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes the content of articles provided via the
$ARGUMENTSpath. - Ingestion points: Reads the article file content and local JSON databases.
- Boundary markers: None present; the agent processes the article text directly to find matching terms.
- Capability inventory: Uses
Read,Edit,Grep, andGlobtools to modify the filesystem. - Sanitization: No explicit sanitization or instruction to ignore embedded commands within the article content is provided.
- [DATA_EXFILTRATION]: The skill performs file system operations that involve traversing to a sibling directory (
../spain-life-guide/) to retrieve configuration data. While this is part of the intended multi-project cross-linking logic, it demonstrates a capability to access files outside the immediate project root. - [COMMAND_EXECUTION]: The skill uses the
Edittool to programmatically modify markdown files based on logic derived from the article content and the link databases. This is a high-privilege operation within the context of the workspace.
Audit Metadata